Design of experiments integrated with neural networks for optimization and predictive modelling of electrode wear of novel Ti-6Al-4V-SiCp composites during die sinking electric discharge machining

Die Sink Electric Discharge Machining is a widely used manufacturing process for shaping hard and electrically conductive materials. This study investigates the effects of various electrode materials such as, Ti-6Al-4V-SiCp, Brass and Copper on the machining performance of AISI 316 l Stainless Steel workpieces using EDM. The methodology involved optimizing parameters such as Electrode Material, Discharge Current, Gap Voltage, Spark Gap, Pulse-on Time, and Pulse-off Time. From the extensive experimantation it was observed that the combination of Ti-6Al-4V-SiCp electrode material, 8Amp Discharge Current, 90 V Gap Voltage, 75 μm Spark Gap, 100 μs Pulse-on Time and 15 μs Pulse-off Time has resulted in lowest electrode eear rate, higher machining time, and low electrode surface roughness ratio. Ti-6Al-4V-SiCp electrodes possess higher hardness and electrical conductivity compared to Brass and Copper Electrodes leading to higher wear resistance against repeated thermal shocks during electric discharge machining operation. Feed Forward Artificial Neural Network is successfully applied to predict the output characteristics of the experimentation with high accuracy of 98.3% (Electrode Wear Rate), 94.6% (Machining Time) and 93.8% (Electrode Surface Roughness Ratio). Further, microstructure analysis concludes that lowest wear is observed in Ti-6Al-4V-SiCp electrodes compared to Brass and Copper electrodes.


Introduction
Manufacturing Industries face a significant challenge in machining of difficult-to-machine materials due to their properties such as brittleness, toughness, chemical reactivity and high hardness.Materials like titanium alloys, hardened steels, nickel-based superalloys, and ceramics pose limitations such as rapid tool wear, poor surface finish, and thermal damage to the workpiece, leading to decreased productivity and increased production costs while being traditionally machined viz.turning, milling, and drilling [1].The difficulty in machining these materials arise from their high strength and hardness, which results in increased cutting forces and temperatures during machining leading to accelerated tool wear and deterioration of surface integrity [2,3].Hard inclusions or abrasive particles present within the material increases tool wear and causes unpredictable tool failure [4].These challenges associated with machining difficult-to-machine materials have given way for nonconventional machining methods such as electric discharge machining (EDM) [5].There are also advanced Powder Mixed Electric Discharge Machining (PMEDM) processes which are capable of machining and coating the samples simultaneously which find a lot of applications in Bio-Medical implants [6].EDM is a thermal machining process that employs electrical discharges to remove material from the workpiece [7,8].In EDM, there is no mechanical contact between the tool and the workpiece, making it suitable for machining materials with high hardness and brittleness compared to the conventional methods [9][10][11].Electric discharge machining is commonly used in micromachining owing to its machining accuracy, dimensional stability and retention of the properties of the workpiece [12].The process begins with the electrode and workpiece positioned close together within the dielectric fluid bath.An electrical potential is applied between the electrode and the workpiece leading to a high-frequency current passage through the gap which ionizes the dielectric fluid, generating a plasma channel.As the plasma channel forms, intense heat is produced, melting and vaporizing material from the workpiece surface [13].Electric discharge die sink machine operates by eroding material from a workpiece through a series of electrical discharges/sparks between the electrode and the workpiece while being submerged in a tank of dielectric fluid.The molten material is flushed away by the stream of dielectric fluid, giving way to precisely machined cavity or a component [14].Figure 1 presents a schematic representation of a die sink electric discharge machine.The electrode is slowly lowered or 'sunk' into the workpiece, gradually shaping the desired geometry through controlled electrical discharges [15].The entire process is highly precise and can achieve intricate shapes with tight tolerances.Several process parameters influence the performance and output quality of electric discharge die sink machining, including discharge current, pulse duration, gap voltage, and flushing rate [16].These parameters directly affect material removal rate, surface finish, and electrode wear.
Longer pulse durations lead to deeper material penetration but increases electrode wear and provides a rougher surface finish.Similarly, higher gap voltage enhances the efficiency of material removal while contributing to a faster electrode wear and increase in surface roughness.Copper and brass electrodes are commonly used in the electric discharge machining (EDM) process due to their excellent electrical conductivity and thermal properties.Copper electrodes or copper plated electrodes are preferred for rough machining applications, while brass electrodes are often used for finishing operations due to their better machinability and lower wear rate [17].Though copper and brass are extensively used as electrodes, they are still susceptible to high wear rates during the EDM process [18].Continuous arcs produced between the electrode and workpiece lead to erosion and deformation of the electrode surface, resulting in increased machining time and rough surface finish.Moreover, as the electrodes wear, the gap between the electrode and workpiece needs to be adjusted periodically to maintain machining accuracy, leading to longer setup times and reduced productivity.Furthermore, surface roughness changes in electrodes after machining can affect the quality and accuracy of subsequent machining operations [19] .The formation of irregularities and micro-cracks on the electrode surface can compromise dimensional accuracy and surface finish, necessitating frequent electrode replacements and additional finishing steps [20].There exists a significant gap in the development of highly efficient and reliable electrodes for EDM applications, particularly in machining difficult-to-machine materials.Traditional copper and brass electrodes have limitations in terms of wear resistance, machinability, and thermal stability, especially when machining difficult to machine materials like titanium alloys and ceramics.To overcome these challenges, extensive research is conducted on advanced materials such as M-Xene phase materials for fabrication of EDM electrodes [21].M-Xene materials are a class of two-dimensional transition metal carbides or nitrides, exhibiting exceptional chemical stability, mechanical strength, and high electrical conductivity [22,23].By utilizing M-Xene-based EDM electrodes, manufacturers can potentially improve machining efficiency, reduce electrode wear, and enhance surface finish [24,25].Electrode wear is a critical factor in electric discharge die sink machining as it directly impacts machining accuracy, dimensional stability, and surface finish.Disproportionate electrode wear leads to dimensional inaccuracies, increased machining time, and poor surface quality, necessitating frequent electrode replacements and compromising productivity.Further, worn electrodes may exhibit irregularities and micro-cracks, further deteriorating machining performance and affecting the overall quality of machined components.Optimization of process parameters in any machining process is crucial for achieving the desired precision, surface finish, and material removal rate.Various methods have been employed to optimize these parameters, each with its own strengths and limitations [26,27].Taguchi's DOE is known for its systematic approach to experimentation, allowing for the simultaneous study of multiple factors with minimal experimental runs.This method provides valuable insights into the main effects and interactions of process parameters.Recent research has shown the potential benefits of combining Taguchi's DOE with Feedforward Artificial Neural Networks (FFANN).Neural Networks excel at learning complex patterns from data, making them well-suited for predicting optimal process parameters based on experimental results.When combined with Taguchi's DOE, Neural Networks can further enhance the optimization process by providing accurate predictions of process parameter effects and interactions.This integrated approach leverages the strengths of both statistical analysis and machine learning, resulting in a powerful optimization technique for EDM process parameters.
The existing literature on Die sink EDM (Electrical Discharge Machining) has primarily focused on optimizing parameters such as material removal rate (MRR), electrode wear rate (EWR), and surface roughness (SR) of the workpiece.While these studies have provided valuable insights into individual aspects of the EDM process, there remains a significant literature gap regarding the simultaneous optimization of multiple input parameters and their impact on electrode performance.One notable aspect of this gap is the lack of comprehensive studies considering the combined influence of electrode material, discharge current, gap voltage, spark gap, pulse on time, and pulse off time on both EWR and Sr While previous research has explored the effect of these parameters on either EWR or SR individually, the holistic understanding of their interplay and collective impact on electrode wear and surface quality is still lacking.Moreover, most of the existing research has predominantly focused on traditional electrode materials such as Brass, Copper, and Graphite.While these materials have been widely utilized in industrial EDM applications, they suffer from limitations in terms of strength and durability, particularly when subjected to prolonged machining operations.Addressing this gap, the current research aims to explore the potential of advanced Titanium-based electrodes, specifically Ti6Al4V-SiCp, in Die sink EDM.Titanium-based materials offer exceptional mechanical properties, including high strength, excellent corrosion resistance, and elevated temperature stability, making them promising candidates for demanding EDM applications.By exploiting the unique characteristics of Ti 3 SiC 2 phase that is formed during Vacuum Sintering of powder metallurgical Ti-6Al-4V-SiCp(15 Wt%) electrodes [28], this study aims to not only optimize the machining process for enhanced EWR and SR performance but also to overcome the limitations associated with conventional electrode materials.Additionally, the incorporation of SiC particles into the Ti6Al4V matrix introduces reinforcement, further enhancing the electrode's mechanical properties and wear resistance [29].By deriving the optimal settings for these parameters using soft computing techniques, such as Taguchi's design of experiments and feed forward artificial neural network (FFANN), this research contributes to the advancement of manufacturing processes.

Methodology
In this experiment, AISI 316 l Stainless Steel is selected as the workpiece and is purchased from Hi-Tech sales Corporation Mangalore.Die sink electric discharge drilling of the workpiece of dimension 200 × 100 × 10 mm is conducted using Sparkonix ZNC 50 A die sink electric discharge machine available at Manipal Institute of Technology, Manipal, India.The composition and properties of the AISI 316 l stainless steel are presented in table 1 and table 2 as provided by the manufacturer.AISI 316 l stainless steel is a popular choice in aerospace, medical and automotive industries due to its high corrosion resistance, hardness and high temperature resistance.AISI 316 stainless steel samples are difficult to machine conventionally because of their high hardness and toughness and are thus suitable for die sink electrical discharge machining.Three electrodes viz.brass, copper and Ti-6Al-4V-SiCp of dimensions 10 × 10 x 60 mm are used in the experimentation.Brass and copper electrodes are purchased from Hi-Tech Sales Corporation Mangalore.Hardness, thermal conductivity and electrical conductivity of Brass, Copper and Ti-6Al-4V-SiCp electrodes are measured in-house and are presented in table 3. Vicker's hardness testing equipment is used to measure the hardness of the samples with a load of 10 kgf.Electrical conductivity is measured at 350 °C using Keith 400 low-temperature electrical resistivity measuring instrument for a sample size of 20 × 10 × 2 mm.Further, thermal conductivity is measured using DLF-2, Waters, Austria-LASER flash thermal analysis system [28].
Ti-6Al-4V-SiCp material is produced using Powder Metallurgy process.Ti-6Al-4V powder of size 100 μm, along with Silicon Carbide powder (100 μm) at 15 Wt% and 3 Wt% PVA (Poly Vinyl Alcohol) is mixed homogenously and compacted using Uniaxial compaction setup.After compacting with a pressure of 4Ton/Sq.inch, the green sample is dried in a muffle furnace at 500 °C, the sample is transferred to a Vacuum Sintering furnace.With a sintering temperature of 1250 °C, sintering time of 3 h, heating and cooling rate of 5 °C min −1 , Ti 3 SiC 2 phase formation has been discovered which contributes towards increase in hardness, thermal and electrical conductivity of the sample [28][29][30].
Input process parameters such as electrode material, discharge current (A), gap voltage (V), spark gap (μm), pulse-on time (μs) and pulse-off time (μs).Electrode wear is calculated by calculating by weighing the electrode before and after machining.Machining time is calculated automatically using the on-board computers of the machine.Electrode Wear Rate (EWR) is then calculated using the following formula.
Surface Roughness of the electrodes is measured using Talysurf Surtronic surface roughness measuring instrument with a cut-off length of 1.00 mm.Three sets of readings are taken for each sample and the average surface roughness is calculated in microns.Surface roughness ratio which can also be termed Surface Roughness Differential is defined as the difference between the surface roughness of the electrode after machining, and is calculated by the following equation.

Surface Roughness Ratio Surface Roughness after machining
Surface Roughness before machining = -Figure 2 presents the Sparkonix ZNC 50 A die sink electric discharge machine.Figures 3 and 4 present the electrodes used in the experimentation along with the microstructure of electrode surfaces before machining.

Design of experiments
Design of experiments is an effective tool used to study the effect of one or more process response variables.Further, it is also an efficient procedure used by various researchers for planning experiments and yield objective conclusions also possible interactions between the process parameters selected.

Taguchi's design of experiments
Taguchi Design of Experiments has been used widely in engineering applications to achieve best levels of a quality characteristic under different conditions.Taguchi's approach to design of experiments is easy to adopt and apply for users with limited knowledge of statistics which resulted wide popularity in engineering and scientific community [31][32][33].In this experimental layout, S/N ratio characteristic 'smaller the better' for has been adopted as given in the equation (1).

Smaller the best characteristic
Where, n is the number of observations and y is the observed data.In this paper, Taguchi L 27 orthogonal array is employed to identify the optimal input process parameters.The levels and factors used for experimentation is shown in table 4.

Feed forward artificial neural network
For applying a Feed Forward Artificial Neural Network to predict the output parameters of the process based on 6 input parameters, the dataset is first preprocessed, and the input parameters are normalized.Subsequently, the data is split into training and testing sets.Then, the neural network architecture is designed and trained using ReLU activation function.The optimal learning algorithm Adam Solver is chosen based on the lowest RMSE, highest R 2 , and lowest MAPE observed in literature survey [34,35].Figure 5 presents the schematic representation of Multi Layered Feed Forward Artificial Neural Network.• Preprocessing the data by normalizing the input parameters and ensuring that the output variables are scaled appropriately.

Model architecture
• Designing a feed forward neural network with multiple hidden layers using TensorFlow and Keras.Experimenting different architectures, including the number of hidden layers, the number of neurons in each layer, and activation functions.

Model training
• Splitting the dataset into training and testing sets.
• Training the model using the training data.
• Monitoring the training process for convergence and adjust hyperparameters as necessary.

Model evaluation
• Evaluating the trained model using appropriate metrics such as Mean Squared Error or R-squared on the testing data.
• Comparing the performance of the model with different activation functions.

Prediction
• Using the trained model to make predictions for new input parameter combinations.

Results and discussions
This section discusses about the results of Electric Discharge Drilling of AISI 316 l SS using different electrodes and input parameters combination followed by the effect of parameters on the output parameters such as, Electrode Wear Rate, Machining Time and Electrode Surface Roughness Ratio (table 5).Further microstructural analysis is presented to provide insights into the electrode behavior assessing the grain boundaries, phases and defects.Finally, the validation of FFANN prediction using experimental results is presented providing comprehensive understanding of the predictive accuracy of the model.

Electrode wear rate
In the series of experiments conducted with Brass electrodes, varying combinations of machining parameters reveal distinct effects on the wear rate.Table 5 and figures 6(a)-(c) present the experimental findings.For instance, when maintaining a constant 90 V gap voltage and 25 μm spark gap, increasing the discharge current from 4 to 12 Amps consistently escalates the wear rate from 331 to 351 μg min −1 .This increase in wear rate can be attributed to the higher thermal energy generated during electrical discharge, causing intensified material removal from the electrodes.At the microscopic level, the higher discharge current results in more intense electrical discharges, leading to localized heating, melting, and vaporization of the electrode material.Similarly, adjusting the gap voltage to 110 V while keeping other parameters constant at 8 Amps discharge current results in a slight decrease in wear rate to 352 μg min −1 .However, this trend is not linear, as seen when both the discharge current and gap voltage are raised to 12 Amps and 110 V, respectively, where the wear rate further increases to 360 μg min −1 .Moreover, variation in the pulse-on time also affects the wear rate, i.e. longer pulseon times generally resulting in lower wear rates.In contrast, experiments with Copper electrodes exhibit a different pattern, wherein variations in discharge current and gap voltage also influence the wear rate.For example, increasing the discharge current from 4 to 12 Amps at a constant 110 V gap voltage leads to a progressive rise in wear rate from 168 to 186 μg min −1 .Similarly, altering the gap voltage to 130 V while maintaining 8 Amps discharge current results in an increase in wear rate from 175 to 194 μg min −1 .Conversely, when the gap voltage is reduced to 90 V with a constant 12 Amps discharge current, the wear rate decreases to 145 μg min −1 .Similarly, experiments with Ti-6Al-4V-SiCp electrodes reveal distinct trends, with variations in discharge current and gap voltage influencing wear rates differently.For instance, increasing the discharge current from 4 to 12 Amps at a constant 130 V gap voltage results in a gradual increase in wear rate from 65 to 76 μg min −1 .Conversely, decreasing the gap voltage to 90 V with a constant 8 Amps discharge current leads to a significant reduction in wear rate to 11 μg min −1 .When subjected to high-voltage electrical discharges, the electrode material undergoes intense thermal and mechanical stresses, leading to material removal through processes such as melting, vaporization, and erosion.For instance, Brass and Copper electrodes experience significant wear due to their lower resistance to the extreme temperatures and mechanical forces generated during EDM [36].This wear process can alter the grain structure of the electrodes, causing grain boundary migration, recrystallization, and even grain refinement in some cases [37].These changes in the grain structure can influence the material's mechanical properties, such as hardness and strength, which in turn affect its wear resistance.The observed increase in wear rate with higher discharge currents in Brass electrodes can be attributed to several specific mechanisms at play.Firstly, the higher discharge current results in an increase in thermal energy generated during electrical discharge, leading to more intense localized heating, melting, and vaporization of the electrode material.This intensified thermal action accelerates material removal from the electrode surface, consequently increasing the wear rate [38].Additionally, the higher discharge current leads to more frequent and intense electrical discharges between the electrode and the workpiece, causing greater mechanical erosion and material displacement.This mechanical action further contributes to the accelerated wear of the Brass electrodes.Moreover, the increased discharge current may also influence the chemical reactions occurring at the electrode surface, such as oxidation or formation of reaction products, which can exacerbate material removal.Therefore, the observed increase in wear rate with higher discharge currents in Brass electrodes can be attributed to a combination of intensified thermal, mechanical, and chemical processes acting on the electrode material during electric discharge machining.In the experiments with Ti-6Al-4V-SiCp electrodes, several factors may explain the gradual increase in wear rate with higher discharge currents while maintaining a constant gap voltage.Firstly, the higher discharge currents result in increased thermal energy input during electrical discharge machining, leading to more intense localized heating and material removal from the electrode surface.This intensified thermal effect contributes to the gradual increase in wear rate observed.Additionally, higher discharge currents can lead to more frequent and intense electrical discharges between the electrode and the workpiece, causing greater mechanical erosion and material displacement, further contributing to wear [39].However, unlike Brass and Copper electrodes, Ti-6Al-4V-SiCp electrodes possess unique properties such as the presence of the Ti 3 SiC 2 phase, which enhances their high-temperature resistance.This means that despite experiencing similar thermal and mechanical stresses, Ti-6Al-4V-SiCp electrodes maintain their structural integrity and mechanical properties better than Brass and Copper electrodes under high discharge currents.Consequently, while the mechanisms underlying wear rate increase with higher discharge currents may be similar across different electrode materials, the unique properties of Ti-6Al-4V-SiCp electrodes afford them greater resistance to wear compared to Brass and Copper counterparts.
From the main effects plot figure 7 it can be deduced that, the best combination of input parameters is, Ti-6Al-4V-SiCp electrode material, 8 Amp Discharge Current, 90 V Gap Voltage, 75 μm Spark Gap, 100μs Pulseon Time and 15μs Pulse-off Time to obtain lowest Electrode Wear Rate since they have lowest Mean of Signal to Noise ratios.
Table 6 presents the ANOVA table with embedded P% (percentage contribution) of each parameter both individually and combined.In the table DF signifies Degrees of Freedom, which represents the number of independent values in a dataset that are free to vary.Seq SS constitutes to sequential sum of squares which indicates the sum of squared deviations of each parameter's effect from the overall mean sequentially.Further, Adj SS is adjusted sum of squares which is Seq SS adjusted for the effect of other parameters in the model.Adj MS (adjusted mean of squares) is the adjusted sum of squares divided by its respective degrees of freedom, representing the average variability attributed to each parameter.Further, F-value is the ratio of variability between groups (parameter effects) to variability within the groups (residual error).It indicates whether the means of different parameter levels are significantly different from each other.Finally, p-value indicates the probability of observing the F-statistic if the null hypothesis were true.A low p-value suggests that the observed effect is unlikely to be due to random chance.Finally, 'S' is the standard error of regression and S = 1.459 suggests the model is accurate.R-Sq is the coefficient of determination and R-Sq = 99.8 suggests that 99.8% of the variability of the dependent variable is explained by the independent variable(s) in the model, indicating a very strong relationship between the variables.The adjusted R-Sq value of 97.4% provides a slightly more conservative estimate of the model's explanatory power, suggesting that the independent variables in the model account for 97.4% of the variability in the dependent variable.From the table we can clearly understand that, Electrode Material is dominant in impacting the Electrode Wear Rate for the given set of range with 86.44% contribution.Further, Gap Voltage has considerable effect with 6.29% followed by Spark Gap with 2.623%.Finally, the combination of Discharge Current and Pulse-on Time has some effect with a percentage contribution of 2.665.The other parameters and their combinations do not provide statistical significance.
Figure 8 presents the SEM images of Ti-6Al-4V-SiCp electrode surface after electric discharge machining under different discharge current (a) 4Amp, (b) 8 Amp, (c) 12 Amps.During EDM, intense localized heating leads to rapid solidification of molten material, resulting in the formation of leaf like structures also known as dendrites.The observed variations in grain boundaries and surface morphology with different gap voltages (130 V, 110 V, and 90 V) during electric discharge machining (EDM) can be attributed to the influence of discharge energy on material removal and solidification kinetics.At higher gap voltages (130 V), the intensified discharge energy facilitates more efficient material removal and rapid solidification, leading to the formation of finer grain structures with reduced surface irregularities.Conversely, lower gap voltages (90 V) result in slower material removal and prolonged solidification, promoting the development of coarser grain boundaries with increased surface roughness.Intermediate gap voltages (110 V) strike a balance between energy input and material removal rates, yielding intermediate grain sizes and surface features.Therefore, the observed variations in grain boundary characteristics across different gap voltages can be elucidated by considering the interplay between discharge energy, material removal dynamics, and solidification processes during EDM.

Machining time
In the experiments conducted to analyze machining time, variations in machining parameters across different electrode materials unveil distinctive trends (figures 9(a)-(c)).For Brass electrodes, manipulating parameters such as discharge current, gap voltage, and pulse-on time leads to discernible changes in machining time.For instance, increasing the discharge current from 4 to 12 Amps consistently reduces the machining time, indicative of higher material removal rates due to intensified thermal energy and discharge densities at the workpiece-electrode interface.Similarly, longer pulse-on times generally result in decreased machining times, reflecting the enhanced efficiency of material disintegration under prolonged electrical discharge conditions.Microscopically, the reduced machining time corresponds to accelerated material removal rates, driven by increased thermal energy and discharge intensities.Conversely, alterations in gap voltage and spark gap dimensions exhibit less pronounced effects on machining time, likely due to the overshadowing influence of discharge current and pulse-on time.In contrast, experiments with copper electrodes reveal similar trends, with variations in discharge current and gap voltage influencing machining times differently.Increasing the discharge current from 4 to 12 Amps consistently decreases machining time, reflecting the heightened material removal efficiency associated with higher energy inputs.Similarly, alterations in gap voltage lead to corresponding changes in machining time, albeit to a lesser extent compared to discharge current variations.Microscopically, the reduced machining time corresponds to accelerated material removal rates, driven by increased thermal energy and discharge intensities.Conversely, variations in Ti-6Al-4V-SiCp electrodes demonstrate analogous trends, underscoring the intricate interplay between machining parameters and machining time across different electrode materials.These observations emphasize the importance of meticulous parameter optimization to strike a balance between productivity and process stability in electric discharge machining processes.
From the main effects plot figure 10 it is found out that, the best combination of input parameters is, Ti-6Al-4V-SiCp electrode material, 8 Amp Discharge Current, 90 V Gap Voltage, 75 μm Spark Gap, 100 μs Pulse-on Time and 15 μs Pulse-off Time to obtain lowest Machining Time since they have lowest Mean of Signal to Noise ratios.
From the table 7 we can clearly understand that, Electrode Material is the most dominant parameter in impacting the Machining Time for the given set of range with 98.82% contribution.Further, Spark Gap has considerable effect with 0.376% followed by Discharge Current with 0.29%.Finally, Gap Voltage has some effect with a contribution of 0.157%.The other parameters and their combinations do not provide statistical significance.

Electrode surface roughness ratio
In the investigation of electrode surface roughness ratio, alterations in machining parameters yield distinctive effects on surface quality across different electrode materials (figures 11(a)-(c)).For Brass electrodes, variations in discharge current, gap voltage, and pulse-on time result in discernible changes in surface roughness ratio.For instance, increasing the discharge current from 4 to 12 Amps generally leads to higher surface roughness ratios, indicative of increased surface irregularities and degradation.This phenomenon is attributed to the heightened  thermal energy and electrochemical reactions during electrical discharge, resulting in more pronounced surface alterations.Microscopically, these alterations arise from complex interactions between electrical, thermal, and chemical phenomena at the electrode surface, including localized heating, melting, and vaporization.Similarly, longer pulse-on times exacerbate surface roughness, as prolonged exposure to electrical discharge intensifies material removal and surface irregularities.Conversely, alterations in gap voltage and spark gap dimensions exhibit less significant effects on surface roughness ratio, likely due to the overriding influence of discharge current and pulse-on time.Conversely, experiments with Copper electrodes reveal analogous trends, with variations in discharge current and gap voltage influencing surface roughness ratio differently.Increasing the discharge current generally leads to higher surface roughness ratios, while alterations in gap voltage result in corresponding changes, albeit to a lesser extent.Microscopically, these alterations stem from similar mechanisms involving localized heating, melting, and vaporization, driven by intense electrical discharges.Conversely, variations in Ti-6Al-4V-SiCp electrodes demonstrate comparable trends, emphasizing the delicate balance required to achieve desired surface finishes while minimizing electrode degradation.These observations underscore the intricate relationship between machining parameters and electrode surface quality, highlighting the need for careful parameter optimization to achieve optimal outcomes in electric discharge machining processes.
From the main effects plot figure 12 it can be deduced that, the best combination of input parameters is, Ti-6Al-4V-SiCp electrode material, 8 Amp Discharge Current, 90 V Gap Voltage, 75 μm Spark Gap, 100 μs Pulseon Time and 15 μs Pulse-off Time to obtain lowest Electrode Surface Roughness Ratio, since they have lowest Mean of Signal to Noise ratios.From the table 8 we can clearly understand that, Electrode Material is dominant in impacting the Surface Roughness Ratio for the given set of range with 70.48% contribution.Further, Gap Voltage has considerable effect with 10.305% followed by Spark Gap with 6.86%.Finally, the combination of Discharge Current and Pulse-on Time has some effect with a contribution of 9.56%.The other parameters and their combinations do not provide statistical significance.

Microscopic analysis
In the analysis of electrode wear post-machining utilizing Die Sink EDM, discernible wear patterns were observed across the three electrode materials examined: Brass, Copper, and Ti6Al4V-SiCp (figure 13).Notably, Brass electrodes exhibited the most pronounced wear, followed by copper, with Ti6Al4V-SiCp displaying the least wear.This variance can be ascribed to the intrinsic properties of each material.Brass, characterized by its relatively lower hardness compared to Copper and Ti6Al4V-SiCp, manifests heightened susceptibility to erosion and wear during the EDM process.
Despite its higher hardness in relation to Brass, Copper experiences considerable wear owing to its diminished resistance to thermal and electrical conductivity when juxtaposed with Ti6Al4V-SiCp.Ti6Al4V-SiCp, a composite material having increased hardness and wear resistance, leads to minimal wear at the electrode edge.Furthermore, throughout EDM operations, microstructural alterations predominantly transpire within the workpiece material, precipitated by the intense thermal flux and abrupt quenching cycles.These transformations encompass the formation of recast layers, heat-affected zones, and grain structure modifications, thereby influencing the material's mechanical attributes and surface integrity.

Validation of FFANN prediction
FFANN, or Feed Forward Artificial Neural Network, is a type of artificial neural network that is commonly used in machine learning applications.The validation of FFANN using experimental results is an important step in determining the accuracy of the model.Table 9 and figures 14(a)-(c)) present the results of FFANN validation.If the error is too high, then it is an indication that the model is not accurate enough and needs to be refined.However, in this case, the error was found to be low, indicating that the FFANN model is accurate enough to be used in predicting the Electrode Wear Rate (98.3% accuracy), Machining Time (94.6% accuracy) and Electrode Surface Roughness Ratio (93.8% accuracy).

Conclusions
Following Conclusions are drawn based on the results of the experimentation.

Electrode wear rate
• The experimental findings across Brass, Copper, and Ti-6Al-4V-SiCp electrodes reveal distinct responses to variations in machining parameters.
• For Brass electrodes, increasing discharge current consistently escalates wear rates due to intensified thermal energy and material removal.However, altering gap voltage and pulse-on time exhibit non-linear effects on wear rates.
• Copper electrodes display similar trends, with variations in discharge current and gap voltage influencing wear rates differently.• Ti-6Al-4V-SiCp electrodes exhibit unique responses, with varying discharge current and gap voltage leading to gradual or significant changes in wear rates.

Machining time
• Brass, Copper, and Ti-6Al-4V-SiCp electrodes showcase distinctive trends in machining time based on variations in discharge current, gap voltage, and pulse-on time.
• Increased discharge current consistently reduces machining time across all electrode materials, reflecting enhanced material removal rates due to intensified thermal energy.• Longer pulse-on times also contribute to decreased machining times, indicative of improved material disintegration efficiency under prolonged electrical discharge conditions.
• Gap voltage and spark gap dimensions exert lesser influence on machining time compared to discharge current and pulse-on time variations.

Electrode surface roughness ratio
• The investigation highlights the nuanced effects of machining parameters on surface quality across Brass, Copper, and Ti-6Al-4V-SiCp electrodes.
• Increased discharge current generally leads to higher surface roughness ratios across all electrode materials, reflecting heightened surface irregularities and degradation.
• Longer pulse-on times exacerbate surface roughness due to intensified material removal and irregularities caused by prolonged exposure to electrical discharge.
• Gap voltage variations exhibit lesser effects on surface roughness compared to discharge current variations.
• Optimal parameter optimization is crucial for achieving desired surface finishes while minimizing electrode degradation in electric discharge machining processes.

Feed forward artificial neural network
• FFANN model is accurate to be used in predicting the Electrode Wear Rate with 98.3% accuracy, Machining Time with 94.6% accuracy and Electrode Surface Roughness Ratio with 93.8% accuracy.

Figure 1 .
Figure 1.Schematic representation of die sink electric discharge machining process.

Figure 3 .
Figure 3. Electrodes used in the experimentation.

Figure 5 .
Figure 5. Schematic representation of multi layered feed forward artificial neural network.

Figure 7 .
Figure 7. Main effects plot for signal-noise ratios.

Figure 10 .
Figure 10.Main effects plot for signal-noise ratios.

Figure 12 .
Figure 12.Main effects plot for signal-noise ratios.

Table 1 .
Composition of AISI 316 l stainless steel.

Table 2 .
Properties of AISI 316 l stainless steel.

Table 3 .
Properties of electrode materials.

Table 6 .
Analysis of variance for SN ratios of electrode wear rate.

Table 7 .
Analysis of variance for SN ratios of machining time.

Table 8 .
Analysis of variance for SN ratios of surface roughness ratio.
Figure 14.Validation of FFANN prediction using experimental results.